Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.gdeltcloud.com/llms.txt

Use this file to discover all available pages before exploring further.

​
What Are Skills?

Skills are MCP resources that provide workflow-pattern guidance. Each skill defines how to approach a class of analytical task: what tools to call, how to sequence them, how to interpret signals, how to structure outputs, and how to cite sources. Skills are served from the MCP server’s skills/tasks directory and loaded at runtime as resources.
Skills complement the research system prompt. The system prompt establishes analyst voice, the four-surface mental model, and citation discipline. A skill provides the deeper workflow for a particular question shape.
Brief production skills are served separately under skills/briefs. The hosted Brief agent preloads the shared tradecraft skill plus one Brief-type skill, then applies the matching briefs.v1 structured output schema from shared contracts.

​
Available Skills

Task skills are oriented around workflow patterns, not vertical topics. Load them on demand based on the question shape.

GDELT Discover-and-Drill

The canonical GDELT Cloud workflow: summarize → search → drill, over-filter recovery, graph traversal recipes for Events, Stories, and Entities.

Macro Finance

Symbol resolution and the ~120-tool macro-finance category — analytics, fundamentals, options chains, sentiment, US macro, FX, crypto, commodities, technical indicators.

Prediction Markets

SEARCH_RELEVANT_MARKETS-first patterns, the proxy-contract pattern for legally-sensitive scenarios, contract-quirk caveats, reading rules_primary and liquidity fields.

Multi-Surface Synthesis

Combining GDELT Cloud + macro-finance + prediction markets to find the divergence; five canonical divergence shapes; weighting rules for conflicting signals.

Brief Output

BLUF / inverted-pyramid analyst brief format with per-evidence-type citation templates and gdelt-event fenced blocks.

Brief Skills

Daily Risk, Crisis SITREP, I&W Watch, and Exposure Monitor skills for the hosted Brief agent.

​
Skill Selection Logic

User RequestSkills
Geopolitical brief on a topicgdelt-discover-and-drill + brief-output
”What’s the market saying / pricing in?”gdelt-discover-and-drill + macro-finance + multi-surface-synthesis + brief-output
Scenario probability with narrative contextgdelt-discover-and-drill + prediction-markets + multi-surface-synthesis + brief-output
Actor exposure / portfolio risk from an eventgdelt-discover-and-drill + macro-finance + multi-surface-synthesis + brief-output
Three-surface synthesis (high-leverage default)All five
Lookup, comparison, or procedural taskNone — answer directly
Generate a Daily Risk / SITREP / I&W Watch / Exposure Monitor Briefbrief-tradecraft + the selected Brief skill
When loading multiple skills, treat gdelt-discover-and-drill as the anchor. It produces the dates, actors, locations, and entities the other skills consume.

​
How Skills Are Loaded

Skills are served as MCP resources with URIs in the format skill://{name}/SKILL.md. 
# Example: load a skill resource via FastMCP client
from fastmcp import Client
from fastmcp.client.auth import BearerAuth

async with Client(
    "https://gdelt-cloud-mcp.fastmcp.app/mcp",
    auth=BearerAuth(token=api_key)
) as client:
    # List available skills
    resources = await client.list_resources()
    skill_resources = [r for r in resources if str(r.uri).startswith("skill://")]

    # Load a specific skill
    skill = await client.read_resource("skill://gdelt-discover-and-drill/SKILL.md")
    print(skill[0].text)
LangGraph or LangChain integrations can preload these resources when initializing the GDELT Cloud MCP server. The hosted GDELT Cloud research agent does this via MCPRemoteSkillsMiddleware, which seeds skill files into the agent’s runtime filesystem so they can be loaded on demand.